The long-term objective of this research program is development of accurate, reliable predictive methods for computation of the flow fields associated with particulate removal in industrial ventilation applications. Improved predictive capabilities are important because the complex interactions between the turbulent gas flow and suspended particulates may pose far greater risks to industrial workers than is currently accounted for by present day design practices. For example, using direct numerical simulation Squires & Eaton (1991a) have demonstrated that turbulence can cause strong preferential concentration of dense particles in regions of low vorticity and high strain rate. This phenomenon can result in instantaneous particle concentrations as much as 25 times the global mean value! To accurately model the interactions between the turbulent flow and suspended particulates existing methods are not satisfactory and more refined approaches are required. Recent developments in subgrid-scale modeling for large-eddy simulation (LES) have made possible much more accurate computation of turbulent flows than has previously been possible. Therefore, the focus of the proposed study is on the development of particle transport models for use in conjunction with large-eddy simulations of turbulence. A new breakthrough in subgrid-scale modeling for LES is adopted for computation of the gas flow in the proposed study. This breakthrough, i.e., dynamic subgrid-scale modeling, permits for the first time LES of turbulent flows without ad hoc adjustment of model coefficients. Particle transport models developed within this framework will provide much more reliable descriptions of turbulence/particle interaction than current methods. The specific aims to be achieved in development of new transport models are: (I) translation of an existing computer code from a high-level vector programming language to Fortran, (2) incorporation of particle tracking capability in the code, (3) formulation of particle transport models, and (4) testing and validation of the models by comparison of results from numerical simulations of turbulent mixing layers to existing experimental measurements. Numerical simulation of particle transport in turbulent mixing layers will provide a rigorous test of the models in flow fields that are representative of those encountered in actual ventilation applications. The proposed research represents the first step towards the development of a new generation of predictive methods which correctly represent turbulence dynamics and particulate transport. The models developed during the course of this study will provide the foundation necessary for accurate prediction of particle transport in contaminant removal problems.